# Post language (the language parameter)

Every endpoint and MCP tool that filters posts uses the same parameter, `language`, to choose which language's posts to query. Values are ISO language codes and are case-insensitive; responses always use the spelling in the table below.

## Accepted values

| Value | Coverage |
| --- | --- |
| `zh` | All Chinese posts: Traditional, Simplified and Cantonese (the default) |
| `zh-Hant` | Traditional Chinese (written Chinese, not Cantonese) |
| `zh-Hans` | Simplified Chinese |
| `yue` | Cantonese: posts written in colloquial Cantonese or mixed with Cantonese |
| `en` | English |
| `ja` | Japanese |
| `ko` | Korean |

- `zh-Hant`, `zh-Hans` and `yue` don't overlap, and together they equal `zh`.
- **If you leave it out, the default is `zh`**, so only Chinese posts come back (except `/post/inspiration`, see below). To get English, Japanese or Korean posts, pass `en`, `ja` or `ko` explicitly. We recommend passing `language` on every call.
- **One language per request.** There's no "all languages" option, and results never mix in posts from other languages. To compare languages, make one call per language.

:::caution[Languages follow the post's writing, not where the author is]
Throk classifies posts by **the language and characters of the post itself**, not by where the author is or their IP. `yue` only covers posts written in colloquial Cantonese or mixed with Cantonese. Posts by Hong Kong users written in standard written Chinese usually land in `zh-Hant`. To capture the full Hong Kong conversation, query `zh`, or query `zh-Hant` and `yue` separately.
:::

## Languages by endpoint

| Endpoint (MCP tool) | Accepted values | Default |
| --- | --- | --- |
| `/post/search` (`search_posts`), `/post/keyword-trend` (`get_keyword_trend`), `/post/hot` (`get_hot_posts`) | All 7 | `zh` |
| `/post/semantic-search` (`semantic_search_posts`) | All 7 | `zh` |
| `/leaderboard/cool-replies` (`get_cool_replies`), `/leaderboard/post-like-growth` (`get_post_like_growth`) | All 7 | `zh` |
| `/leaderboard/follower-growth` (`get_follower_growth`) | All 7 | `zh` |
| `/stats/follower-ladder` (`get_follower_ladder`), `/stats/like-ladder` (`get_like_ladder`), `/stats/power-hours` (`get_power_hours`) | All 7 | `zh` |
| `/post/inspiration` (`get_inspiration`) | All 7 | Not set: Chinese, Japanese and Korean |
| [Webhook](/en/api/webhooks/) rules | `zh`, `zh-Hant`, `zh-Hans`, `yue`, `ja`, `ko` | `zh` |

- Leaderboards, ladders and power hours are **calculated separately for each language**: for example, the `ja` follower ladder only looks at authors of Japanese posts, and numbers are never combined across languages.
- If you leave out `language`, `/post/inspiration` (`get_inspiration`) returns high-engagement Chinese, Japanese and Korean posts, not English ones. Pass `en` for English posts.

## Excluding Cantonese: `exclude_hk`

With `language=zh`, `/post/search`, `/post/hot`, `/post/keyword-trend` and `/post/semantic-search` accept `exclude_hk=true` to exclude Cantonese (`yue`) posts, which gives the same result as `zh-Hant` plus `zh-Hans`. Other languages don't need this parameter.

## How keywords are matched

Keyword matching differs by language: Chinese and Japanese match character by character, Korean matches space-separated words, and English posts (`en`) are case-insensitive and match word forms such as singular and plural together. See [search rules](/en/api/guides/search-rules/).